[Performance] Can onnxruntime delegate the entire process to another framework when using other providers?

[yht183]

Describe the issue

For example, when using the MIGraphXExecutionProvider, the execution frame also records some intermediate variable values of MIGraphX. Is it possible to provide an end-to-end method where onnxruntime only handles input and output data, and the other provider handles everything else?

To reproduce

when I run
ort_session = onnxruntime.InferenceSession(onnx_model_path, providers=["MIGraphXExecutionProvider"], provider_options=provider_options)
I found that the all_values_ in the executionFrame class records many intermediate variables, even when using other providers for inference execution. I wonder if there is a way to perform end-to-end execution when using other providers?

Urgency

No response

Platform

Linux

OS Version

ubuntu2004

ONNX Runtime Installation

Built from Source

ONNX Runtime Version or Commit ID

v1.15.0

ONNX Runtime API

Python

Architecture

X64

Execution Provider

MIGraphX

Execution Provider Library Version

No response

Model File

No response

Is this a quantized model?

No